问题
It's possible, with cabal, to set up a continuous build that records test successes/failures in a format many CI systems will accept with a command like:
cabal test '--test-option=--jxml=test-results/$test-suite.xml'
The important part here is that $test-suite is replaced with the name of the test, so that different tests put their results into different files.
When I use stack, all tests get literally the option --jxml=test-results/$test-suite.xml passed to them, so the end result is that the tests overwrite each others' results.
Is there any way to run all my tests with stack so that I can tell each test suite a different place to write their results?
I'd even accept some stack command that parsed the cabal file for me and told me what test suites there are, because then I could script a loop in bash calling each test one at a time.
回答1:
I'd even accept some stack command that parsed the cabal file for me and told me what test suites there are, because then I could script a loop in bash calling each test one at a time.
If you're willing to stoop to that, stack ide targets will return a list of targets, from which you can do some bashing to get the list of test suites. Something like this:
stack ide targets 2>&1 |
while IFS=: read pkg type suite; do
[[ $type = test ]] && stack test ":$suite" --test-arguments="--jxml=test-results/$suite.xml"
done
回答2:
stack test --test-arguments="--jxml=test-results/$test-suite.xml"
来源：https://stackoverflow.com/questions/55957477/get-stack-to-pass-test-suite-name-as-part-of-test-arguments